Information Technology Reference
In-Depth Information
This distinction is close to the distinction between 'control for prevention' and 'control
for detection' used by (Bons
et al.
, 1998). However, they mainly refer to legal aspects of
contracts, and in general to violations. Our distinction is related to the general theory of action
(the function of control actions) and delegation, and it is more general.
The first form/finality of control (
kindA
) prevents not only violations (in case of norms,
commitments, or contracts) but also missed execution or mistakes (also in weak delegation
where there are no obligations at all).
The second form/finality (
kindB
) is not only for sanctions or claims, but for timely inter-
vening and preventing additional damages, or remedying and correcting. 'Detection' is just a
means; the real aim is intervention for safety, enforcement or compensation.
12
Moreover, an
effect (and a function/aim) of the second form of control can also be to prevent violation; this
happens when the controlled agent knows or believes - before or during his performance -
that there will be 'control for detection' and he worries about this (sanctions, reputation, lack
of autonomy, etc.).
7.1.8 Filling the Gap between Doing/Action and Achieving/Results
Let's put the problem in another perspective. As we said, trust is the background for delegation
and reliance i.e., to 'trust' as a decision and an action; and it is instrumental to the satisfaction
of some goal. Thus the trust in
Y
(sufficient for delegation) implies the trust that
g
(the goal
for which
X
counts on
Y
) will be achieved.
Given these two components or two logical steps scenario, we can say that the first kind
of control is pointing to, is impinging on the first step (trust in
Y
) and is aimed at increasing
it; while the second kind of control is mainly pointing to the second step and is aimed at
increasing it, by confirming the achievement of
g
also in case of (partial) defaillance of
Y
.
In this way the control (monitoring plus intervention) complements the trust in
Y
which
would be insufficient for achieving
g
, and for delegating; this additional assurance (the possi-
bility to correct work in progress of
Y
's activity) makes
X
possible to delegate to
Y
the goal
g
.
In fact, in this case
X
is not only counting on
Y
,but
X
counts on a potential multi-agent plan
that includes her own possible actions.
As we can see from the formula (7.5a) in Section 7.1.7 the important thing is that
Y believes
that the control holds, and not if it really holds.
13
For example,
X
could not trust
Y
enough
and communicate to him the control: this event modifies
Y
's mind and
X
's judgment about
Y's
trustworthiness. Thus, in trust-reliance, without the possibility of intervention for correction
and adjustment, there is only one possible way to achieve
g
, and one activity (
Y
's activity) on
which
X
bets (Figure 7.7).
Meanwhile, if there is control for correction/adjustment, the achievement of
g
is committed
to
Y
's action plus
X
's possible action (intervention),
X
bets on this combination (Figure 7.8).
A very similar complementing or remedying role are guarantees, protections and assurance.
I do not trust the action enough, and I put protections in place to be sure about the desired
12
Different kinds of delegation (weak, mild, strong: see Section 2.9.1) allow for specific functions of this control.
There will be neither compensation nor sanctions in weak delegation (no agreement at all), while there will be
intervention for remedy.
13
This is the actual power of the Gods.
Search WWH ::
Custom Search